Choose the right machine learning algorithm in Amazon SageMaker
summary
Amazon SageMaker exposes 17 built-in algorithms and demonstrates an image classification pipeline using ResNet-152 with transfer learning on a public chest X‑ray set of 112,000 images, training on two p3.16xlarge instances (each with eight V100 GPUs), using mixed precision and 10 epochs, then one‑line deployment to a realtime endpoint. For tabular problems, XGBoost is explained as gradient‑boosted decision trees with many hyperparameters, distributed HPO, feature store integration at Nexi, and production inference via batch transform or HTTPS endpoints.
AI-written summary · about
chapters (6)
- 0:00Introduction to SageMaker algorithms
- 5:02Demo of image classification
- 10:38Choosing algorithms for tabular data
- 13:49Understanding XGBoost
- 21:00Nexi’s XGBoost use case
- 27:48Summary and resources
related talks
19mDeveloping large-scale machine learning algorithms on Amazon SageMaker
19mUsing Amazon SageMaker to develop and deploy ML models
20mIndustrial-Scale Machine Learning
32mBuild, Train, and Deploy Machine Learning Models on AWS
33mBuild, train, and deploy ML models using Amazon SageMaker
44mPragmatic Machine Learning in the Cloud
34mBuild, train and deploy Machine Learning models with Amazon SageMaker
45mDeploying ML Models to Production with Amazon SageMaker